llm-intelligence/scripts/import_phase2_data_test.go

//go:build llm_script

package main

import "testing"

func TestEnrichBaiduModelMetadataUsesSpecificFamilyRules(t *testing.T) {
	cases := []struct {
		modelID         string
		wantReleaseDate string
		wantSourceURL   string
		wantConfidence  string
		wantSourceKind  string
	}{
		{
			modelID:         "baidu-ernie-5.0",
			wantReleaseDate: "2026-01-22",
			wantSourceURL:   "https://cloud.baidu.com/news/news_eacd0f0b-0ca3-4963-aec8-5e6b9ebef9ba",
			wantConfidence:  "official_primary",
			wantSourceKind:  "official_announcement",
		},
		{
			modelID:         "baidu-ernie-x1.1",
			wantReleaseDate: "2025-09-09",
			wantSourceURL:   "https://cloud.baidu.com/news/news_be713ff4-8477-4852-88f1-9cc56c406d6a",
			wantConfidence:  "official_primary",
			wantSourceKind:  "official_announcement",
		},
		{
			modelID:         "baidu-ernie-4.5-turbo-128k",
			wantReleaseDate: "2025-04-25",
			wantSourceURL:   "https://cloud.baidu.com/article/3887765",
			wantConfidence:  "official_primary",
			wantSourceKind:  "official_announcement",
		},
		{
			modelID:         "baidu-ernie-x1-turbo-32k",
			wantReleaseDate: "2025-04-25",
			wantSourceURL:   "https://cloud.baidu.com/article/3887765",
			wantConfidence:  "official_primary",
			wantSourceKind:  "official_announcement",
		},
		{
			modelID:         "baidu-ernie-4.5-8k",
			wantReleaseDate: "2025-03-16",
			wantSourceURL:   "https://cloud.baidu.com/article/3835921",
			wantConfidence:  "official_primary",
			wantSourceKind:  "official_announcement",
		},
		{
			modelID:         "baidu-ernie-x1-8k",
			wantReleaseDate: "2025-03-16",
			wantSourceURL:   "https://cloud.baidu.com/article/3835921",
			wantConfidence:  "official_primary",
			wantSourceKind:  "official_announcement",
		},
		{
			modelID:         "baidu-ernie-character",
			wantReleaseDate: "2024-03-22",
			wantSourceURL:   "https://cloud.baidu.com/news/news_667c065f-0bd7-475d-98c2-901763d0ee77",
			wantConfidence:  "official_primary",
			wantSourceKind:  "official_announcement",
		},
	}

	for _, tc := range cases {
		enriched := enrichBaiduModelMetadata(ModelPricing{
			ModelID:   tc.modelID,
			SourceURL: "https://cloud.baidu.com/doc/qianfan/s/wmh4sv6ya",
		})

		if enriched.ReleaseDate != tc.wantReleaseDate {
			t.Fatalf("%s release date = %q, want %q", tc.modelID, enriched.ReleaseDate, tc.wantReleaseDate)
		}
		if enriched.ModelSourceURL != tc.wantSourceURL {
			t.Fatalf("%s source url = %q, want %q", tc.modelID, enriched.ModelSourceURL, tc.wantSourceURL)
		}
		if enriched.DateConfidence != tc.wantConfidence {
			t.Fatalf("%s date confidence = %q, want %q", tc.modelID, enriched.DateConfidence, tc.wantConfidence)
		}
		if enriched.DateSourceKind != tc.wantSourceKind {
			t.Fatalf("%s date source kind = %q, want %q", tc.modelID, enriched.DateSourceKind, tc.wantSourceKind)
		}
	}
}

func TestEnrichBaiduModelMetadataFallsBackToPricingSource(t *testing.T) {
	enriched := enrichBaiduModelMetadata(ModelPricing{
		ModelID:   "baidu-unknown-preview",
		SourceURL: "https://cloud.baidu.com/doc/qianfan/s/wmh4sv6ya",
	})

	if enriched.ReleaseDate != "" {
		t.Fatalf("unexpected release date: %q", enriched.ReleaseDate)
	}
	if enriched.ModelSourceURL != "https://cloud.baidu.com/doc/qianfan/s/wmh4sv6ya" {
		t.Fatalf("model source url = %q, want pricing source fallback", enriched.ModelSourceURL)
	}
	if enriched.DateConfidence != "unknown" || enriched.DateSourceKind != "unknown" {
		t.Fatalf("unexpected fallback date metadata: confidence=%q kind=%q", enriched.DateConfidence, enriched.DateSourceKind)
	}
}

func TestEnrichBaiduModelMetadataSupportsSourceOnlyRules(t *testing.T) {
	enriched := enrichBaiduModelMetadata(ModelPricing{
		ModelID:   "baidu-ernie-4.5-turbo-vl",
		SourceURL: "https://cloud.baidu.com/doc/qianfan/s/wmh4sv6ya",
	})

	if enriched.ReleaseDate != "" {
		t.Fatalf("unexpected release date: %q", enriched.ReleaseDate)
	}
	if enriched.ModelSourceURL != "https://cloud.baidu.com/product/wenxinworkshop.html" {
		t.Fatalf("model source url = %q, want product source", enriched.ModelSourceURL)
	}
	if enriched.DateConfidence != "unknown" || enriched.DateSourceKind != "official_product_page" {
		t.Fatalf("unexpected source-only metadata: confidence=%q kind=%q", enriched.DateConfidence, enriched.DateSourceKind)
	}
}

func TestEnrichBaiduModelMetadataSupportsOfficialPrimaryRules(t *testing.T) {
	enriched := enrichBaiduModelMetadata(ModelPricing{
		ModelID:   "baidu-ernie-5.1",
		SourceURL: "https://cloud.baidu.com/doc/qianfan/s/wmh4sv6ya",
	})

	if enriched.ReleaseDate != "2026-05-09" {
		t.Fatalf("release date = %q, want %q", enriched.ReleaseDate, "2026-05-09")
	}
	if enriched.ModelSourceURL != "https://ernie.baidu.com/blog/posts/ernie-5.1-0508-release/" {
		t.Fatalf("model source url = %q, want official announcement", enriched.ModelSourceURL)
	}
	if enriched.DateConfidence != "official_primary" || enriched.DateSourceKind != "official_announcement" {
		t.Fatalf("unexpected official metadata: confidence=%q kind=%q", enriched.DateConfidence, enriched.DateSourceKind)
	}
}

func TestEnrichBaiduModelMetadataSupportsCatalogBackfillRules(t *testing.T) {
	enriched := enrichBaiduModelMetadata(ModelPricing{
		ModelID:   "baidu-qianfan-vl-70b",
		SourceURL: "https://cloud.baidu.com/doc/qianfan/s/wmh4sv6ya",
	})

	if enriched.ReleaseDate != "" {
		t.Fatalf("unexpected release date: %q", enriched.ReleaseDate)
	}
	if enriched.DateConfidence != "unknown" || enriched.DateSourceKind != "catalog_backfill" {
		t.Fatalf("unexpected catalog metadata: confidence=%q kind=%q", enriched.DateConfidence, enriched.DateSourceKind)
	}
}

func TestBaiduReleaseDateValueReturnsNilForUnknownDate(t *testing.T) {
	if value := releaseDateValue(""); value != nil {
		t.Fatalf("blank release date should stay nil, got %v", value)
	}
	if value := releaseDateValue("invalid"); value != nil {
		t.Fatalf("invalid release date should stay nil, got %v", value)
	}
}